18 source Fail rows lack instant-off.
The stored Fail remains visible. The draft leaves those rows unscored rather than filling in a value or creating a false grade.
Case study · First pass
From CP station records to a focused review plan.
A draft cathodic protection review framework that makes the source outcome, proposed scoring, and unresolved questions easy to see together.
Draft for specialist reviewAggregate findings only
The challenge
The data had answers—and questions.
The supplied workbook brought together station inventory, survey readings, calculated values, and the company's stored Pass/Fail result. A useful review needed to preserve that result while showing where the evidence was complete, missing, or open to interpretation.
Project aim
Give a CP specialist a better starting point.
Turn the source into a consistent, traceable review queue. Show the reason behind each provisional grade or flag. Keep AI output advisory and leave criteria approval with the CP3 and company.
What the first pass found
A clear view of the whole set.
These are aggregate figures from the working snapshot. The station-level workbook, identifiers, readings, and locations are not published on this site.
155station/location rows across 22 system labels
81provisional condition grades
74unscored due to insufficient core evidence
111rows with at least one rule-based review flag
Source result, preserved
The supplied workbook records 88 Fail, 7 Pass, and 60 blank outcomes. The draft does not overwrite them or turn a condition grade into a compliance verdict.
Provisional grade, separate
Among the rows with enough evidence for a grade: 1 A, 41 B, 31 C, and 8 D. No row received a provisional F. These bands are discussion aids under a proposed rubric.
Where review matters
The flags point to decisions, not blame.
Several patterns need interpretation before the rubric can guide field decisions.
31 rows need a closer look.
These source Fail rows show at least a 100 mV difference between recorded Off and Depolarized readings. The workbook's formula adds another depolarized-potential condition. The CP3 needs to confirm the intended criterion and whether the readings are comparable.
9 isolation entries disagree.
Matched entries differ between inventory and current-status views. The review queue keeps that disagreement visible until the authoritative entry is confirmed.
Reading-pair question: the consolidated sheet dates Off to June 2023 and Depolarized to July 2024. The intended pairing needs confirmation. A later-year On/Off column label also needs review.
The deliverable
A working review package.
01
Transparent grades
Seven documented factors and explicit coverage rules produce a provisional condition grade only where the minimum evidence is present.
02
Focused review queue
Rows with missing inputs, low coverage, or source differences carry a specific review reason, alongside the original company outcome.
03
Advisory AI signal
TypeSafe Jev made a structured second pass across all 155 de-identified rows, returning urgency, probabilities, and confidence. Low-confidence cases remain visible; Jev does not change the grade or source result.
The detailed five-tab spreadsheet is shared privately with authorized recipients. Read the public method and rubric.
Next step
Expert review makes this useful.
A CP3 and the company need to confirm the applicable criteria, reading pairs, labels, and rubric weights. The proposed score curves can then be calibrated against representative specialist-ranked cases before operational use.